March 2017 doc IEEE 802 11 170274 r

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Further Evaluations on Single Carrier Waveforms Date: 2017 -03 -01 Authors: Submission Slide 1 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Abstract • In [1] and [2], we introduced a DFT-spread operation to 802. 11 ay and evaluated the throughput performance in comparison to the ones with OFDM and SC waveforms • In this contribution, • We provide more clarifications on the receiver structures for SC waveforms in [2] • We discuss why SC receivers degrades the PER performance • Additional simulation results are also provided Submission Slide 2 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Background (1/2) • In 802. 11 ad [3], both OFDM and SC waveforms are adopted and similar waveforms are expected to be adopted for 802. 11 ay [4] • In 802. 11 ad, the numerologies for SC and OFDM are chosen such that both SC and OFDM can operate at the same nominal sample rate • The nominal sample rate for a single channel is 2. 64 GHz Submission Slide 3 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Background (2/2) • 802. 11 ad OFDM TX and RX do not need to employ extra upsampling and downsampling operations [3] • On the other hand, a typical SC TX and RX for 802. 11 ay/ad may employ an upsampling, filtering, and a downsampling operations to match the sampling rate • SC receiver operates on a block where the channel’s impact is a circular convolution to exploit 1 -tap freq. domain equalization Pilot 336 16 OFDM TX and RX: M a p 0 P / S IDFT (512) DAC 448 Data P / 64 S GI Submission Filter DAC RF RF S / P ADC DFT (512) 352 E Q 0 Data SC TX and RX: RF RF Slide 4 DAC Filter Data S / P 448 DFT (512) E Q IDFT (512) 64 Alphan Sahin (Inter. Digital) Data GI

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Equivalent Representation of SC (1/2) • Upsampling, filtering, and downsampling operations for single carrier can expressed in frequency. For example, the equivalent operations for single channel are given as follows: • • • Upsampling: DFT of 768, repeat in frequency, IDFT of 1536 [7, p 238] Downsampling: DFT of 1536, combine in frequency, IDFT of 512 [7, p 235] Filtering: Windowing in frequency domain (corresponds to 1536 filter coefficients) Cancels each other Submission DFT (512) E Q IDFT (512) Filter coef. Data IDFT 448 (512) GI 64 DFT (512) Cancels each other Slide 5 Output of DFT S / P Output of DFT 768 (768) 768 Filter SC receiver DFT (768) Filter coef. Weighted Comb. 512 Filter coef. Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Equivalent Representation of SC (2/2) • The equivalent frequency domain representation of 11 ay/ad SC receiver includes for single channel • A 768 -DFT operation in front and a weighted combination before the equalization DFT (512) E Q 448 Data IDFT (512) 64 448 DFT (768) RS Amplitude SC receiver Complex coefficients frequency Channel frequency response Submission Slide 6 E Q Weighted Comb. Output of DFT S / P Output of DFT Filter • This receiver is optimal in AWGN with a matched filter • However, one can show that this receiver is suboptimal in fading channel IDFT (512) 64 Data RS Non-coherent additions before equalization! Filter coef. Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Simulation Assumptions AP PAA STA Fixed location and orientation Random location and orientation Submission Slide 7 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 BER Results without PA Impairment SC receiver performs ~ 1 d. B worse than others in the channel b) 11 ay channel a) AWGN • In AWGN, all waveforms are aligned with Gaussian BER bound • In fading channel, conventional SC receiver degrades BER performance approximately 1 d. B due to aforementioned noncoherent combinations Submission Slide 8 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 BER Results with PA Impairment 1 d. B SC receiver loses its PAPR advantage for low SNR SC is ~1 d. B better than OFDM b) 11 ay channel a) AWGN • In AWGN, SC and DFT-spread OFDM outperform OFDM due to their low PAPRs • In 11 ay channel, SC receiver does not capture the PAPR advantage as compared to DFT-spread OFDM for low SNR Submission Slide 9 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 PER Results in 11 ay Channel ~1 d. B ~0. 8 d. B a) Without PA impairment (13/16 coding rate) b) With PA impairment (13/16 coding rate) • With or without PA impairment, the SC receiver is ~1 d. B worse than DFT-spread OFDM • With PA impairment, both SC and OFDM are ~0. 8 d. B worse than DFT-spread OFDM for 13/16 coding rate. Throughput result is provided in Appendix II • The results for ½ coding rate is provided in Appendix III and IV Submission Slide 10 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Conclusion • In this contribution, we discuss the upsamplingfiltering downsampling operation at the SC receiver • We showed that the upsampling-filtering downsampling operation at the SC receiver is equivalent to a weighted combination in frequency domain, which is not optimal in fading channel • Simulations show that • The SC receiver degrades the FER performance as compared to DFT-spread OFDM • With high coding rate and PA impairments, DFT-spread OFDM is superior to OFDM Submission Slide 11 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 References [1] “On the Single Carrier Waveforms for 11 ay”, IEEE 802. 1116/01455 r 0 [2] “Performance Evaluation of Multi-DFT-spread OFDM for 802. 11 ay ”, IEEE 802. 11 -17/00048 r 0 [3] “Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 3: Enhancements for Very High Throughput in the 60 GHz Band, ” IEEE Std 802. 11 ad 2012, pp. 1– 628, Dec. 2012. [4] “Specification Framework for TGay, ” IEEE 802. 11 -15/01358 r 6] [5] “Channel Models for IEEE 802. 11 ay, ” IEEE 802. 11 -15/1150 r 7 [6] “ 11 ay evaluation methodology, ” IEEE 802. 11 -16/866 r 4 [7] K. R. Rao at al. , “Fast Fourier Transform - Algorithms and Applications”, Signals and Communication Tech. , Springer, 2010 Submission Slide 12 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Appendix I - Waveforms 0 P 64 x 4 / S GI DFT-S-W-OFDM * DFT-S-OFDM * SC 448 x 4 Data RRC 299 x 4 GI GI 299 x 4 DFT 53 x 4 RF DAC IDFT (2048) RF 0 DAC P / S DFT (2048) 352 x 4 Data EQ 0 RRC RF RF ADC S / P DFT (2048) 352 x 4 EQ IDFT (2048) EQ IDFT 448 x 4 Data 0 DAC IDFT (2048) 0 ADC RF P / S ADC S / P DFT 53 x 4 Data RF Data DAC CP 256 RF RF ADC Amplitude EQ 16 x 4 CP+ 256 Weighted Comb. Pilot P / S IDFT (2048) a p Phase EQ 336 x 4 M Window OFDM Data IDFT 0 *Numerology is selected such that it has the same throughput of SC and OFDM approximately Submission Slide 13 Alphan Sahin (Inter. Digital) Data

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Appendix II - Throughput Results in 11 ay Channel (16 QAM, 13/16) ~1 d. B ~0. 8 d. B a) Without PA impairment (13/16 coding rate) Submission Slide 14 b) With PA impairment (13/16 coding rate) Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Appendix III - PER Results in 11 ay Channel (16 QAM, 1/2) ~1 d. B ~0. 9 d. B a) Without PA impairment (½ coding rate) Submission b) With PA impairment (½ coding rate) Slide 15 Alphan Sahin (Inter. Digital)

March 2017 doc. : IEEE 802. 11 -17/0274 r 0 Appendix IV - Throughput Results in 11 ay Channel (16 QAM, 1/2) ~1 d. B ~0. 9 d. B a) Without PA impairment (½ coding rate) Submission b) With PA impairment (½ coding rate) Slide 16 Alphan Sahin (Inter. Digital)